On Wed, May 10, 2006, Michael Crowley wrote:
> Would the pb code developers please take a look at line 544 in pb_direct.f
> uips=1.0d0/rips/rinv
>
> rinv is undefined at this point. Why has this not caused an error up till
> now? The compiler on alpha-compaq warns about this, but apparently this
> code never is used, at least not by the test cases.
There is no reason to use IPS with poisson-boltzmann, so that all of
the IPS code sections (marked with WXW) should be removed, in my opinion.
This should be certainly be done (assuming everyone agrees) in amber10; I
don't think a patch to amber9 is needed, since it *should* be impossible to
reach that section of code: it would require setting ips>0 and igb=10, which
is already flagged as an illegal combination of input parameters in mdread().
Ray: I am still concerned that we keep having to make all these changes
twice, once in src/pbsa and once in src/sander. The code in these two
directories started out the same, but is now diverging. Is there any good way
to avoid all of this duplication?
...thx...dac
Received on Sun May 14 2006 - 06:07:06 PDT
